English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This paper addresses the challenge of 3D human pose estimation from a singlecolor image. Despite the general success of the end-to-end learning paradigm,top performing approaches employ a two-step solution consisting of aConvolutional Network (ConvNet) for 2D joint localization and a subsequentoptimization step to recover 3D pose. In this paper, we identify therepresentation of 3D pose as a critical issue with current ConvNet approachesand make two important contributions towards validating the value of end-to-endlearning for this task. First, we propose a fine discretization of the 3D spacearound the subject and train a ConvNet to predict per voxel likelihoods foreach joint. This creates a natural representation for 3D pose and greatlyimproves performance over the direct regression of joint coordinates. Second,to further improve upon initial estimates, we employ a coarse-to-fineprediction scheme. This step addresses the large dimensionality increase andenables iterative refinement and repeated processing of the image features. Theproposed approach outperforms all state-of-the-art methods on standardbenchmarks achieving a relative error reduction greater than 30% on average.Additionally, we investigate using our volumetric representation in a relatedarchitecture which is suboptimal compared to our end-to-end approach, but is ofpractical interest, since it enables training when no image with corresponding3D groundtruth is available, and allows us to present compelling results forin-the-wild images.

Coarse-to-Fine Volumetric Prediction for Single-Image 3D Human Pose